Re: [PATCH v2 2/6] gdbserver: Add PID parameter to linux_get_auxv and linux_get_hwcap

[Thiago Jung Bauermann <thiago.bauermann@linaro.org>]

Simon Marchi <simark@simark.ca> writes:

> On 11/28/22 06:50, Luis Machado via Gdb-patches wrote:
>> Hi,
>> 
>> On 11/26/22 02:04, Thiago Jung Bauermann wrote:
>>> This patch doesn't change gdbserver behaviour, but after later changes are
>>> made it avoids a null pointer dereference when HWCAP needs to be obtained
>>> for a specific process while current_thread is nullptr.
>>>
>>> Fixing linux_read_auxv, linux_get_hwcap and linux_get_hwcap2 to take a PID
>>> parameter seems more correct than setting current_thread in one particular
>>> code path.
>> 
>> I'm wondering if passing the pid is needed at all in gdbserver for the purposes of
>> fetching the hwcap entries from auxv.
>> 
>> While current thread can be nullptr, I suppose current_process will never be nullptr if
>> we have a valid inferior. And the auxv entries are per-process rather than per-thread.
>> 
>> There is a bit of a corner case when we have extended-remote without a live process and
>> we need to fetch the hwcap bits to determine a particular feature to report via
>> qSupported. But I suppose this is not what this change is trying to address, is it?
>> 
>> If not, it may be the case that we don't need these changes, and we can just use
>> current_process.
>
> Even if we could just use current_process, this patch is going in the
> general direction I like, which is to reduce referring to the global
> state everywhere, try to fetch it only at the entry points and then pass
> the necessary context down by parameters.

Thanks. As I looked into how to fix the problem I ran into, I thought
this was a nice cleanup that would solve my problem and couldn't resist
doing it. :-)

-- 
Thiago